1. Introduction
Injury is the fifth leading cause of mortality and morbidity among adults in Iowa. Work-related, or occupational injuries, constitute a substantial proportion of the injury burden in the US. The Bureau of Labor Statistics reports about 4.9 million occupational injuries annually in the US and over 50,000 injuries in Iowa. The aims of this study were to assess: 1) the burden of work-related injuries using an algorithm based on analysis of the external cause of injury (i.e., E-codes) combined with injury registry variables; 2) the magnitude of disability following an occupational injury by assessing Disability Adjusted Life Years and Discharge to Long Term Care; and 3) the burden of short-term disability one year after discharge from a Level I trauma center.
2. Methods/Approach
This research consisted of two observational studies of Iowa cases, ages 18-64, reported to the Iowa State Trauma Registry (STR) from January 1st 2007 to December 31st 2010. A retrospective cohort design was used to assess differences in mortality, length of stays, discharge disposition, disability risk and Disability-Adjusted Life Years (DALYs), and their associated risk factors, such as demographic characteristics, nature and severity of injury, pre-hospital and in hospital trauma care (i.e. transport, resuscitation, vascular and airway access, sedative and paralytic drug usage).
A prospective follow-up study a cohort of cases discharged from the University of Iowa Resource Trauma Center was used to assess risk factors associated with short-term disabilities one year after discharge. The EuroQol-5 Dimension Questionnaire (EQ-5D) was mailed to the cohort cases, alive one year after discharge, to assess their overall health status and quality of life. The algorithm classified the study population into occupational, “Work-Likely” (WL), and non-work cases. Work-likely was defined based on work-related activities without pay, informal status or self-employed. The registry cases were matched to 2007-2011 death certificates to identify those who died after discharge and to estimate their survival time. Machine learning methods – logistic regression and 10-fold cross validation were used to validate the algorithm. The survival time from injury to death was assessed using Kaplan Meier and Cox regression modeling. The Generalized Linear Modeling, including multinomial regression, was used to analyze the mean length of stay, the risk of discharge to long term care, DALYs and disability risk.
3. Results
From 2007 to 2010, there were individuals (ages 18-64) admitted (average 5,614 per year) as trauma cases to hospitals in Iowa. Based on the algorithm, 3,115 (14.0%) were classified as occupational, 847 (3.8%) as WL, and 18,454 (82.2%) were classified as non-work cases. There were notable differences in demographics, farm exposure, and rural residence. The 10-fold cross validation showed a 20% misclassification rate for occupational and 30% for WL. The area under the curve (AUC) of the receiver-operating characteristics (ROC) was measured at 0.66, which is indicative of poor discriminating effect. Overall, occupational and WL cases had better outcomes than non-work cases; they had lower mean lengths of stay and better survival rates, as detected by Kaplan-Meier and Cox regression models. WL had lower survival rates on the Kaplan Meier estimates but the Cox regression contrast statement didn't find any difference in survival between occupational and WL 30 days and one year after discharge. Multinomial regression showed major differences in the risk of discharge to long-term care (LTC) or acute care compared to discharge home between occupational, WL, and non-work cases WL cases had less risk of being discharged to LTC compared to non-work cases. There were no differences observed between occupational and non-work cases. When stratified by occupational status, the predictors of being discharged to Long term care or acute care were different for occupational, WL and non-work cases. For WL, care in Level I&II , injury type, triage mechanism, first ER systolic blood pressure were no longer good predictors of discharge to LTC compared to occupational or non-work cases. Mean DALYs were lower in the occupational (mean= 4.8; 95% CI: 4.7-4.8) and WL (mean 4.4; 95% CI: 4.4-4.7) cases than the non-work cases (mean= 5.2; 95% CI: 5.1-5.2). However, when all other risk factors were accounted for, the occupational cases had a 10% reduction in mean DALYs, and WL a 20% reduction in DALYs compared to non-work cases. When the disability was assessed separately by occupational status, the risk factors associated with disability were completely different. For WL cases, only injury location and ISS were significantly associated with DALYs. Conversely for occupational and non-work cases, injury type, coma, pre-hospital management (i.e. airway, paralytic drugs), and cause of injury were significantly associated with DALYs. The one-year follow-up questionnaire administered to 156 trauma survivors resulted in 72 (46%) valid responses. Of those who responded, 58 (81%) were occupational and 14 (19%) WL cases. Overall, from the EQ-5Dresults, 46% of the respondents reported a disability. There were no major differences in the prevalence of disability between occupational and WL injury cases. However, occupational injury cases were more likely to receive rehabilitation services.
4. Conclusions
This study demonstrated the utility of using trauma registry data in epidemiologic research to study occupational injury using a unique algorithm to include informal or self-employed workers. It identified a neglected group of workers subject to occupational injury and subsequent disability.
| Identifer | oai:union.ndltd.org:uiowa.edu/oai:ir.uiowa.edu:etd-6414 |
| Date | 01 May 2016 |
| Creators | Diallo, Ousmane |
| Contributors | Torner, James C. |
| Publisher | University of Iowa |
| Source Sets | University of Iowa |
| Language | English |
| Detected Language | English |
| Type | dissertation |
| Format | application/pdf |
| Source | Theses and Dissertations |
| Rights | Copyright 2016 Ousmane Diallo |
Page generated in 0.0688 seconds